The setup
Hurricane Marie peaked as a Category 2 far offshore and never threatened landfall; its south swell reached Southern California over the first weekend of September 2026, under high-surf and coastal-flood advisories. On September 6 and 7 we were on the beach at Newport with cameras and a drone, capturing timestamped observations at multiple spots while logging the model's values for the same hours.
What the model said, and what the ocean did
At 15:00 local time the model reported essentially identical conditions for all eight Newport spots in the dataset: 1.40 m of swell at 10.3 s from 184–186°. On the water at the same hour, River Jetties was stacked with lines of whitewater on a red-flag day; Blackies, 3 km away, was essentially flat — glassy, families wading; and The Wedge was breaking well overhead with a rescue ski standing by.
The model was not wrong about the offshore sea state. It was silent about the only thing a person on the sand needs: which of two beaches minutes apart is safe, flat, or firing. That resolution gap — offshore model to beach-scale decision — is the gap AYON Wave's spot records exist to close, and it can only be closed with observation.
What changed because we were there
Three spot records were verified on site, and one was corrected: a secondary swell window that the draft data claimed for Blackies was deleted, because the ocean said no. Every new spot record in the app ships unverified until a timestamped field observation reviews it — the count of verified records is a public, auditable measure of the product's honesty, and on that weekend it grew by three.
This is research. Estimates are models, not guarantees; a model is never an observation, and a machine estimate is never “confirmed” — only a human reviewing it in the field makes it verified. Nothing on this page is an operational-capability claim, a safety promise, or a prediction of a specific event at a specific place.
References & data
- 1.NHC — Hurricane MARIE (EP13) advisory archive, 2026 — All 30 advisory cycles, 1 Sep 0900 UTC to 8 Sep 1500 UTC 2026 (forecast/public advisories, discussions, wind-speed probabilities). This is the storm record the swell timing is checked against.
- 2.NDBC station 46222 — San Pedro, CA (Scripps Waverider buoy) — The MEASUREMENT baseline used in the field log: 6 Sep 2026 02:26 UTC (19:26 PDT) WVHT 1.2 m at DPD 13 s from 167°, water 23.1 °C; evening trend 1.3–1.4 m at 11–13 s from 160–173°. Historical data is downloadable from the station page.
- 3.Open-Meteo marine + wind at 33.5934, −117.8815 (MODEL) — The modeled baseline logged for the same hour: 19:00 PDT total 1.7 m at 12.7 s from 186°, swell 1.6 m at 11.2 s from 184°; wind 8–9 kt gusting 12 from 353–050° (offshore for that stretch). The article's identical-values observation for eight Newport spots at 15:00 comes from this source.
- 4.AYON Ecosystem ground-truth log — The Wedge, 2026-09-06 — Internal file docs/ground-truth-days/2026-09-06-wedge-mary.md: witness account (video 19:06 PDT), the buoy and model rows above, and five still montages plus three video frames. The stills and frames are on file and not published here; they will be linked only after the people in them and the author approve publication. No National Weather Service advisory text is quoted because we did not archive the advisory itself — only the storm archive above is cited.